4028
Biochemistry 2000, 39, 4028-4036
Oxidase Reaction of Cytochrome cd1 from Paracoccus pantotrophus† Alrik Koppenho¨fer,‡,§ Richard H. Little,‡,| David J. Lowe,⊥ Stuart J. Ferguson,*,§ and Nicholas J. Watmough*,| Department of Biochemistry, UniVersity of Oxford, South Parks Road, Oxford OX1 3QU, United Kingdom, Centre for Metalloprotein Spectroscopy and Biology, School of Biological Sciences, UniVersity of East Anglia, Norfolk NR4 7TJ, United Kingdom, and Biological Chemistry Department, John Innes Centre, Colney Lane, Norwich NR4 7UH, United Kingdom ReceiVed August 16, 1999; ReVised Manuscript ReceiVed December 2, 1999
ABSTRACT: Cytochrome cd1 (cd1NIR) from Paracoccus pantotrophus, which is both a nitrite reductase and an oxidase, was reduced by ascorbate plus hexaamineruthenium(III) chloride on a relatively slow time scale (hours required for complete reduction). Visible absorption spectroscopy showed that mixing of ascorbate-reduced enzyme with oxygen at pH ) 6.0 resulted in the rapid oxidation of both types of heme center in the enzyme with a linear dependence on oxygen concentration. Subsequent changes on a longer time scale reflected the formation and decay of partially reduced oxygen species bound to the d1 heme iron. Parallel freeze-quench experiments allowed the X-band electron paramagnetic resonance (EPR) spectrum of the enzyme to be recorded at various times after mixing with oxygen. On the same millisecond time scale that simultaneous oxidation of both heme centers was seen in the optical experiments, two new EPR signals were observed. Both of these are assigned to oxidized heme c and resemble signals from the cytochrome c domain of a “semi-apo” form of the enzyme for which histidine/methionine coordination was demonstrated spectroscopically. These observations suggests that structural changes take around the heme c center that lead to either histidine/methionine axial ligation or a different stereochemistry of bishistidine axial ligation than that found in the as prepared enzyme. At this stage in the reaction no EPR signal could be ascribed to Fe(III) d1 heme. Rather, a radical species, which is tentatively assigned to an amino acid radical proximal to the d1 heme iron in the Fe(IV)-oxo state, was seen. The kinetics of decay of this radical species match the generation of a new form of the Fe(III) d1 heme, probably representing an OH--bound species. This sequence of events is interpreted in terms of a concerted two-electron reduction of oxygen to bound peroxide, which is immediately cleaved to yield water and an Fe(IV)-oxo species plus the radical. Two electrons from ascorbate are subsequently transferred to the d1 heme active site via heme c to reduce both the radical and the Fe(IV)-oxo species to Fe(III)-OH- for completion of a catalytic cycle.
Cytochrome cd1 is a respiratory nitrite reductase enzyme for which the reaction product is nitric oxide. The enzyme is a dimer, with each monomer comprising two distinct domains, one of which contains a c-type cytochrome, the other a d1 heme. The d1 heme is a ferric dioxoisobacteriochlorin (2, 3), a prosthetic group found exclusively in this class of bacterial enzyme. Structural studies confirm that the d1 heme is the site of nitrite reduction (4), while the function of the c-type cytochrome is to mediate electron transfer from electron donor proteins to the active site. The crystal structures of the enzymes from both Paracoccus pantotrophus1 and Pseudomonas aeruginosa have recently been solved to k-1 there will never be a significant accumulation of intermediate X* and the formation of Y will be rate-limited by the formation of the bimolecular complex (X*). Under these circumstances the dependence upon [O2] of the measured rate constant (kobs) for the overall reaction over the range 92-600 µM will appear to be linear. In
contrast, for k2 < k-1 the dependence of kobs upon substrate concentration would be described by a rectangular hyperbole even in this limited range of substrate concentrations. Simulated plots of kobs versus [O2] (not shown) reveal that the minimum value of k2 required to maintain an apparently linear relationship under the experimental conditions is of the order of 150 s-1 when the value of Keq is constrained to 500 M-1. On longer time scales (>1 s) we observed a slow increase in absorbance at 550 nm that suggested that rereduction of heme c was taking place (Figure 3C). This was surprising given the sluggish reduction of the oxidized enzyme at pH 6.0, which took place over a period of approximately 2 h (Figure 1A). To investigate this further we recorded spectra at 0.65, 10.5, 30.2, and 60.3 s after mixing of reduced cd1NIR with dioxygen (Figure 3D). The spectrum recorded after 0.65 s [Figure 3D (s)] is very close to the calculated spectrum of species Z [Figure 3B (- - -)] and represents the product of the initial reaction of the reduced enzyme with dioxygen. However, in the spectra recorded at 10.5 s [Figure 3D (- - -)] and 30.2 s (Figure 3D (‚‚‚)] there is an increase in absorbance at 550 nm, confirming that heme c becomes partially rereduced. In addition there are changes in the spectrum of the d1 heme in the visible region with a decrease in absorbance at 650 nm and an increase in absorbance at 720 nm. The spectrum recorded at 60.3 s showed no further changes (data not shown). Intermediates of Dioxygen Reduction. The question as to the chemical nature of the intermediates of oxygen reduction that are bound to the d1 heme iron during turnover was
4032 Biochemistry, Vol. 39, No. 14, 2000
Koppenho¨fer et al.
FIGURE 3: Spectrokinetic analysis of the reaction of reduced cd1NIR with oxygen. The reaction was carried out as described in the legend to Figure 2 except that the stopped-flow spectrophotometer was fitted with a diode-array detector. Spectra recorded over the first 0.5 s are shown in panel A. The data were fitted by global analysis to the three-component model as described in the text, in which the spectrum of X was constrained to that of the fully reduced enzyme. The spectrum of X (s) together with the calculated spectra of the reaction intermediates Y (‚‚‚) and Z (- - -) are shown in panel B. To observe the progress of the reaction on longer time scales, a second experiment was done in which 200 spectra were recorded over 60 s. Panel C shows the time courses of the absorbance changes at 550 nm (heme c) on a logarithmic time scale. Panel D shows spectra recorded at 0.65 s (s), 10.2 s (- - -), and 30.2 s (‚‚‚).
addressed by use of freeze-quench EPR (Figure 4). On the same time scale as the simultaneous oxidation of both heme centers was seen in the optical experiments, we observed new signals arising from oxidized heme c (see later), but no signal that could be ascribed to ferric d1 heme. However, we noticed the formation of an organic radical species (g ) approx. 2.002, Figure 4) whose X-band EPR spectrum is shown in Figure 5. This radical species was readily apparent at 25 ms (Figure 4), reached its maximum intensity at 77 ms, and then decayed (Figure 4). There was no evidence of this radical species in the EPR spectra of fully reduced cd1NIR either before or after mixing with oxygen-free buffer (not shown). The radical spectrum (Figure 5) is most intense at low temperature (8 K) and relatively high power (120 µW), which suggests that the radical lies close to a second paramagnet, presumably a metal and therefore probably one of the two hemes in cd1NIR. The fine structure of the radical spectrum (Figure 5), which clearly shows splitting into five lines, is similar to that reported for a number of tyrosine radicals in metalloproteins (21, 22). All these features of the g ) 2.0033 signal are inconsistent with the possibility of it arising from an ascorbate radical, which has a distinct two-line EPR spectrum, in the bulk phase. After 77 ms the signal associated with the organic radical was fully developed (Figure 4), but neither heme was reduced (Figure 2) and there was no EPR signal that could be attributed to an isolated Fe(III) d1 heme. This would suggest that the radical species is a feature of intermediate Y. The
lack of an Fe(III) d1 heme EPR spectrum, and the unusual relaxation properties of the organic radical, could be explained by intermediate Y containing d1 heme in the Fe(IV)-oxo state. The disappearance of this radical species correlates well with the formation of a rhombic EPR spectrum (gx ) 2.45, gy ) 2.22, gz ) 1.86), which increased in intensity from 77 to 500 ms (Figure 4A). The closely spaced g-values of this rhombic species are very similar, but not identical, to the components of the spectrum arising from oxidized low-spin heme d1 with histidine and tyrosine as axial ligands (gx ) 2.52, gy ) 2.19, gz ) 1.84) described by Cheesman et al. (23) (Figure 4B). These differences lead us to propose that the novel rhombic spectrum observed in the freeze-quench experiments arises from an Fe(III) low-spin d1 heme with histidine and OH- as axial ligands. There are two reasons for this assignment. First, such a compressed rhombic EPR spectrum is characteristic of an Fe(III) lowspin heme such as ferric dioxoisobacteriochlorin in which the ground-state electronic configuration is formally (dxzdyz)4(dxy)1 (23, 24). Second, this species would be the expected product of the one-electron reduction of an Fe(IV)-oxo d1 heme. Nature of the Axial Ligands to Heme c. The freezequench EPR studies give insight into not only the events at the d1 heme but also the nature of the axial ligands to heme c. The X-ray structure of the oxidized enzyme as prepared shows the c-type heme to have an unusual bis-histidine coordination, in which the planes of the imidazole groups are perpendicular to each other. This arrangement gives rise
Oxidase Reaction of P. pantotrophus Cytochrome cd1
Biochemistry, Vol. 39, No. 14, 2000 4033
FIGURE 4: Progress of the reaction of reduced cd1NIR with dioxygen, monitored by X-band EPR spectroscopy. Reaction conditions were essentially as described in the legend to Figure 2 except that the concentration of cd1NIR was 60 µM and that, after mixing, samples were rapidly quenched by freezing at the time points shown according to the method of Bray et al. (15). Panel A shows spectra of samples frozen after 25, 77, and 202 ms. Panel B shows time points of 500 and 1000 ms (note the gain in panel B is 2.5 times that in panel A). Spectra were recorded at 10 K under the following conditions: microwave frequency 9.64 GHz, microwave power 2 mW, and modulation amplitude 1 mT.
FIGURE 5: X-band EPR spectrum of the transient radical species seen during the reaction of reduced cd1NIR with dioxygen. The EPR spectrum of the radical species trapped after 25 ms (Figure 4, panel A) was recorded at 5.5 K under the following conditions: microwave frequency 9.6404 GHz, microwave power 120 µW, and modulation amplitude 0.3 mT.
to a high gmax-type EPR signal of which only the gz component can be seen at g ) 3.05 (23) (Figure 6B). Within 25 ms of the reaction of the reduced enzyme with dioxygen, new EPR signals are seen. The major signals at g ) 2.93 and g ) 2.26 are presumed to be the gz and gy components of a rhombic trio that persists for at least 5 s (Figure 4 and
data not shown). These values are close to those observed for the same components of the EPR spectrum arising from the heme c center of P. aeruginosa cd1NIR (23), where the axial ligands are known from the crystal structure (6). A second minor EPR signal from the holoenzyme is also apparent in early time points (Figure 4A) as a broad shoulder centered at g ) 3.4. This broad signal, which we have yet to assign, progressively disappears such that after 500 ms only the dominant g ) 2.93 signal remains. The EPR spectrum of the “semi-apo” form of the P. pantotrophus enzyme also contains two major features at g ) 2.92 and g ) 2.26. In this form of the enzyme, the d1 heme has been chemically removed, leaving heme c as a single prosthetic group. “Semi-apo” cd1NIR has an electronic absorption spectrum that contains a weak transition at 695 nm (Figure 6A), the position and intensity of which is characteristic of low-spin ferric hemes possessing at least one sulfur-containing ligand. Cytochrome c (histidine/ methionine), bacterioferritin (bis-methionine) and cytochromes P-450 (cysteine/H2O) all give rise to such intensity (25-27). The structures of the oxidized and reduced forms of the holoenzyme confirm that His-69 is adjacent to two cysteine residues (at positions 65 and 68) that covalently bind the vinyl groups of the heme. We assume that this ligand does not vary and that the sulfur ligand to the c-heme in the “semi-apo” enzyme is provided by Met-106 as in the reduced holoenzyme (4). Our interpretation of the g ) 2.93 and g ) 2.26 signals that appear following oxygenation of the fully reduced
4034 Biochemistry, Vol. 39, No. 14, 2000
FIGURE 6: Electronic absorption and X-band EPR spectra of the “semi-apo” form of cd1NIR. Panel A shows the room-temperature electronic absorption spectra of the oxidized (- - -) and dithionitereduced states (s) of the “semi-apo” form of cd1NIR. The protein concentration was 5 µM in 50 mM potassium phosphate buffer, pH 7.0. The absorbance between 600 and 800 nm is shown on a 50-fold expanded scale to illustrate the absorbance feature at 695 nm. Panel B shows the X-band EPR spectra of the fully oxidized states of the holo and “semi-apo” forms of cd1NIR. The concentration of holo cd1NIR was 700 µM and that of the “semi-apo” cd1NIR was 470 µm. Both samples were prepared in 50 mM potassium phosphate, pH 7.0. The spectra were recorded at 8 K under the following conditions: microwave frequency 9.44 GHz (holo cd1NIR) or 9.57 GHz (“semi-apo” cd1NIR), microwave power 2.0 mW, and modulation amplitude 1 mT.
enzyme is that His-69 and Met-106 persist as axial ligands to heme c, at least on the time scale of our experiments, and that it is this form of heme c that gives rise to the new EPR signals. However, strictly speaking we cannot exclude that these two signals arise from a different arrangement of bishistidine-coordinated heme c than that observed in the asprepared enzyme. Work is currently in progress in our laboratories to try and discriminate between these two possibilities using magnetic circular dichroism (MCD) spectroscopy. DISCUSSION The reaction of cd1NIR from P. aeruginosa with its various substrates has been studied by stopped-flow techniques over several years. The best-known result to emerge from this
Koppenho¨fer et al. work is that the rate of electron transfer between the c- and d1-type hemes is slow (kobs > 5 s-1) compared to the likely rate of protonation of an oxygen atom on nitrite bound to the d1 heme iron. The latter event generates water and a nitrosyl group bound to the ferrous state of the heme. The latter can be regarded as equivalent to an Fe(III)-NO complex from which the NO seems to dissociate slowly because a dead-end Fe(II)-NO complex forms even though the rate of electron transfer to the d1 heme is slow (8). This slow rate of internal electron transfer from the c to the d1 heme irons in cd1NIRs is a matter of considerable interest, because in the absence of any structural changes, it is hard to reconcile with a distance of just 12 Å between the edges of the heme planes. The cd1NIR from P. aeruginosa also catalyzes an oxidase reaction, during which the rate of electron transfer from the c to d1 heme is much faster, on the order of 100 s-1 (11). It has been uncertain to what extent this interesting, but often overlooked, result can be extrapolated to the P. pantotrophus cd1NIR. This is especially true because it is now clear that the enzymes from the two sources have a number of quite distinct structural features. To address this issue we have used a combination of time-resolved spectroscopies to study the oxidase reaction of the P. pantotrophus cd1NIR and show that, in common with the P. aeruginosa enzyme, interheme electron transfer occurs on the millisecond time scale when reduced P. pantotrophus cd1NIR reacts with dioxygen. The optical changes seen on mixing the fully reduced P. pantotrophus cd1NIR with dioxygen are reminiscent of those reported by Greenwood et al. (11) for the same reaction in the P. aeruginosa enzyme. These authors monitored the reaction at 665 nm, although the absorbance maximum of P. aeruginosa cd1NIR is in fact 650 nm, and observed a kinetic profile similar to that we report for the P. pantotrophus enzyme at 640 nm. In both enzymes the fast phase of the reaction, which is essentially complete in about 100 ms, involves bleaching of the absorbance associated with the reduced d1 heme and the concomitant oxidation of the c-heme. Although Greenwood et al. noted that interheme electron transfer was rapid (k > 100 s-1), it is not a feature of their kinetic model (eq 1 of ref 11). Consequently, the reported affinty of O2 for the reduced d1 heme (Ka ) 1 × 104 M-1), which is based upon an assumption of simple and reversible binding, may be an overestimate. Our preferred model for the P. pantotrophus enzyme involves the weak binding of oxygen to the Fe(II) d1 heme and fast hemeheme electron transfer to reduce the bound oxygen to peroxide. A similar strategy is used by respiratory hemecopper oxidases to trap kinetically dioxygen (28). Once the initial phase of oxidation is complete, we find no evidence of the type of low-spin Fe(III) d1 heme with peroxide as the sixth axial ligand. Instead we see the formation of an organic radical that we have tentatively assigned to a tyrosine residue close to the d1 heme iron. The distal ligand to the Fe(III) d1 heme in oxidized cd1NIR is Tyr-25 (5), but this is not necessarily the tyrosine that forms the radical since there is structural evidence that this residue moves away from the d1 heme upon reduction (4). Instead we favor the participation of the conserved tyrosine at position 263, which is hydrogen-bonded to the d1 heme macrocycle and only 6.5 Å away from the d1 heme iron, which might account for the relaxation properties of the
Oxidase Reaction of P. pantotrophus Cytochrome cd1 radical that we observe. Confirmation of the participation of a tyrosine radical in the oxygen cleavage chemistry would be significant. The rapid reduction of molecular oxygen to an Fe(IV)-oxo species plus water requires three electrons. Peroxidases (in which the substrate, peroxide, is already two-electron reduced) and oxidases use a variety of strategies to supply the third electron. These include oxidation of either the porphyrin macrocyle in horseradish peroxidase (HRP) (29) and ascorbate oxidase (30), a tryptophan residue in yeast cytochrome c peroxidase (CCP) (31), and a covalently modified tyrosine residue in cytochrome aa3 (32). In each of these enzymes it is possible to generate the radical species in vitro for spectroscopic examination, but in each case turnover is sufficiently rapid to preclude direct observation of the participation of the radical in catalysis. The sluggishness of the cd1NIR oxidase reaction allows observation of not only the formation and rereduction of the radical but also, in parallel spectrokinetic experiments, the formation and decay of an Fe(IV)-oxo species at the d1 heme. The rereduction of cd1NIR that occurs subsequent to its initial oxidation (Figure 2) is important for understanding subsequent events that take place at the d1 heme on longer time scales (Figures 3 and 4). If the reduction of cd1NIR by ascorbate/HARC were always slow, as it is initially for the oxidized (as-prepared) enzyme, then rereduction of the enzyme would not be expected over the time course shown in Figure 4, in which case only two electrons would be available on each polypeptide chain. Since the structure shows that transfer of electrons between redox centers on different polypeptides in a dimer is unlikely because of distance considerations, then under these circumstances dioxygen bound to either monomer could not be fully reduced to two molecules of water. However, because we observe the formation and subsequent decay of the radical species, additional electrons must have been supplied to the d1 heme center on the time scale of our experiments. If reduction of dioxygen is complete (i.e., four electrons are supplied to each polypeptide chain) then the final product formed could either be two waters or one water plus one bound hydroxide. This would be entirely consistent with our interpretation of the freeze-quench EPR spectra recorded at time points longer than 0.2 s. We have observed that reduction of the “semi-apo” enzyme, which has histidine/ methionine coordination of the oxidized c-type center, by ascorbate is much faster than that of the holoenzyme. This would be consistent with the form of the holoenzyme generated immediately after oxidation of its c-type center being able to react faster with ascorbate, and thus gain additional electrons for the reduction of the initial species formed by the reaction of reduced cd1NIR with oxygen, than the initial as-prepared state of the enzyme, which has bishistidine coordination. The time-resolved EPR spectra reported here (Figure 4) show the existence of an oxidized heme c center in P. pantotrophus cd1NIR, the axial ligands of which are different from those observed in either the crystal or solution structures of the oxidized enzyme as prepared (5, 23). The signals at g ) 2.93 and g ) 2.26 (Figure 4) are similar to signals reported in both the P. aeruginosa holoenzyme (33) and the P. pantotrophus “semi-apo” enzyme (Figure 6), both of which have histidine/methionine axial coordination. This strongly
Biochemistry, Vol. 39, No. 14, 2000 4035 suggests that we have observed the same axial coordination for heme c in the reoxidized P. pantotrophus holoenzyme. It is important to realize that this observation implies that at least during turnover with oxygen, bis-histidine coordination of heme c depends on Tyr-25 being bound to the d1 heme iron. When this cannot occur, as in the “semi-apo” enzyme, the heme c switches to histidine/methionine axial coordination, even in the oxidized state. The intermediate species Y and Z analyzed in this paper are each proposed to have an intermediate derived from O2 bound to the d1 heme iron. Thus one can expect this to result in the displacement of Tyr-25 in species Y and Z, an observation that would correlate with our proposal of histidine/methionine axial coordination for heme c in these species. ACKNOWLEDGMENT We thank the following: Professor M. T. Wilson, University of Essex, for his suggestions concerning the reaction mechanism, Professor A. J. Thomson and Dr. M. R. Cheesman, School of Chemical Sciences, UEA, for helpful discussions concerning the assignment of the EPR spectra, and Dr. D. J. Richardson, School of Biological Sciences, UEA, for his critical reading of the manuscript. REFERENCES 1. Rainey, F. A., Kelly, D. P., Stackebrandt, E., Burghardt, J., Hiraishi, A., Katayama, Y., and Wood, A. P. (1999) Int. J. Sys. Bacteriol. 49, 645-651. 2. Chang, C. K. (1985) J. Biol. Chem. 260, 9520-9522. 3. Chang, C. K., and Wu, W. (1985) J. Biol. Chem. 261, 85938596. 4. Williams, P. A., Fu¨lo¨p, V., Garman, E. F., Saunders, N. F. W., Ferguson, S. J., and Hajdu, J. (1997) Nature 389, 406412. 5. Fu¨lo¨p, V., Moir, J., Ferguson, S., and Hajdu, J. (1995) Cell 81, 369-77. 6. Nurizzo, D., Silvestrini, M. C., Mathieu, M., Cutruzzola, F., Bourgeois, D., Fu¨lo¨p, V., Hajdu, J., Brunori, M., Tegoni, M., and Cambillau, C. (1997) Structure 5, 1157 1171. 7. Parr, S. R., Barber, D., Greenwood, C., and Brunori, M. (1977) Biochem. J. 167, 447-455. 8. Silvestrini, M. C., Tordi, M. G., Musci, G., and Brunori, M. (1990) J. Biol. Chem. 265, 11783-11787. 9. Timkovich, R., and Robinson, M. K. (1979) Biochem. Biophys. Res. Commun. 88, 649-55. 10. Wharton, D. C., and Gibson, Q. H. (1976) Biochim. Biophys. Acta 430, 445-453. 11. Greenwood, C., Barber, D., Parr, S. R., Antonini, E., Brunori, M., and Colosimo, A. (1978) Biochem. J. 173, 11-17. 12. Kobayashi, K., Koppenho¨fer, A., Ferguson, S. J., and Tagawa, S. (1997) Biochemistry 36, 13611-13616. 13. Moir, J. W. B., Baratta, D., Richardson, D. J., and Ferguson, S. J. (1993) Eur. J. Biochem. 212, 377-85. 14. Berks, B. C., Richardson, D. J., Robinson, C., Reilly, A., Aplin, R. T., and Ferguson, S. J. (1994) Eur. J. Biochem. 220, 117124. 15. Bray, R. C., Lowe, D. J., Capeille`re-Blandin, C., and Fielden, E. M. (1973) Biochem. Soc. Trans. 1, 1067-1072. 16. Kaye, G. W. C., and Laby, T. H. (1966) p 153, Tables of Physical and Chemical Constants. Longmans, London. 17. Parr, S. R., Wilson, M. T., and Greenwood, C. (1974) Biochem. J. 139, 273-276. 18. Parr, S. R., Wilson, M. T., and Greenwood, C. (1975) Biochem. J. 151, 51-59. 19. Barber, D., Parr, S. R., and Greenwood, C. (1976) Biochem. J. 157, 431-438.
4036 Biochemistry, Vol. 39, No. 14, 2000 20. Bagshaw, C. R., Eccleston, J. F., Eckstein, F., Goody, R. S., Gutfreund, H., and Trentham, D. R. (1974) Biochem. J. 141, 351-64. 21. Ivancich, A., Jouve, H. M., Sartor, B., and Gaillard, J. (1997) Biochemistry 36, 9356-64. 22. Rigby, S. E., Nugent, J. H., and O’Malley, P. J. (1994) Biochemistry 33, 1734-42. 23. Cheesman, M. R., Ferguson, S. J., Moir, J. W. B., Richardson, D. J., Zumft, W. G., and Thomson, A. J. (1997) Biochemistry 36, 16267-16276. 24. Cheesman, M. R., and Walker, F. A. (1996) J. Am. Chem. Soc. 118, 7373-7380. 25. Cheesman, M. R., Kadir, F. H. A., Albasseet, J., Almassad, F., Farrar, J., Greenwood, C., Thomson, A. J., and Moore, G. R. (1992) Biochem. J. 286, 361-367. 26. Eaton, W. A., and Hochstrasser, R. M. (1967) J. Chem. Phys. 46, 2533-2539.
Koppenho¨fer et al. 27. McKnight, J., Cheesman, M. R., Thomson, A. J., Miles, J. S., and Munro, A. W. (1993) Eur. J. Biochem. 213, 683-687. 28. Verkhovsky, M. I., Morgan, J. E., Puustinen, A., and Wikstro¨m, M. (1996) Nature 380, 268-270. 29. Dolphin, D., Forman, A., Borg, D. C., Fajer, J., and Felton, R. H. (1971) Proc. Natl. Acad. Sci. U.S.A. 68, 614-8. 30. Bonagura, C. A., Sundaramoorthy, M., Pappa, H. S., Patterson, W. R., and Poulos, T. L. (1996) Biochemistry 35, 6107-15. 31. Sivaraja, M., Goodin, D. B., Smith, M., and Hoffman, B. M. (1989) Science 245, 738-40. 32. MacMillan, F., Kannt, A., Behr, J., Prisner, T., and Michel, H. (1999) Biochemistry 38, 9179-9184. 33. Walsh, T. A., Johnson, M. K., Greenwood, C., Barber, D., Springall, J. P., and Thomson, A. J. (1979) Biochem. J. 177, 29-39. BI991912K
